diff --git a/expose.h b/expose.h
index 029a2f0d2..326053c35 100644
--- a/expose.h
+++ b/expose.h
@@ -54,7 +54,6 @@ struct load_model_inputs
     const int cublas_info = 0;
     const char * vulkan_info = nullptr;
     const int blasbatchsize = 512;
-    const int debugmode = 0;
     const int forceversion = 0;
     const int gpulayers = 0;
     const float rope_freq_scale = 1.0f;
@@ -64,6 +63,8 @@ struct load_model_inputs
     const float tensor_split[tensor_split_max] = {};
     const int quant_k = 0;
     const int quant_v = 0;
+    const bool quiet = false;
+    const int debugmode = 0;
 };
 struct generation_inputs
 {
@@ -97,7 +98,6 @@ struct generation_inputs
     const bool stream_sse = false;
     const char * grammar = nullptr;
     const bool grammar_retain_state = false;
-    const bool quiet = false;
     const float dynatemp_range = 0.0f;
     const float dynatemp_exponent = 1.0f;
     const float smoothing_factor = 0.0f;
@@ -157,6 +157,7 @@ struct sd_load_model_inputs
     const char * vae_filename = nullptr;
     const char * lora_filename = nullptr;
     const float lora_multiplier = 1.0f;
+    const bool quiet = false;
     const int debugmode = 0;
 };
 struct sd_generation_inputs
@@ -172,7 +173,6 @@ struct sd_generation_inputs
     const int seed = 0;
     const char * sample_method = nullptr;
     const int clip_skip = -1;
-    const bool quiet = false;
 };
 struct sd_generation_outputs
 {
@@ -187,6 +187,7 @@ struct whisper_load_model_inputs
     const int clblast_info = 0;
     const int cublas_info = 0;
     const char * vulkan_info = nullptr;
+    const bool quiet = false;
     const int debugmode = 0;
 };
 struct whisper_generation_inputs
@@ -195,7 +196,6 @@ struct whisper_generation_inputs
     const char * audio_data = nullptr;
     const bool suppress_non_speech = false;
     const char * langcode = nullptr;
-    const bool quiet = false;
 };
 struct whisper_generation_outputs
 {
@@ -214,6 +214,7 @@ struct tts_load_model_inputs
     const char * vulkan_info = nullptr;
     const int gpulayers = 0;
     const bool flash_attention = false;
+    const bool quiet = false;
     const int debugmode = 0;
 };
 struct tts_generation_inputs
@@ -221,7 +222,6 @@ struct tts_generation_inputs
     const char * prompt = nullptr;
     const int speaker_seed = 0;
     const int audio_seed = 0;
-    const bool quiet = false;
     const bool nocache = false;
 };
 struct tts_generation_outputs
diff --git a/gpttype_adapter.cpp b/gpttype_adapter.cpp
index 03c1c4e50..811b2f20e 100644
--- a/gpttype_adapter.cpp
+++ b/gpttype_adapter.cpp
@@ -106,7 +106,7 @@ static kcpp_params * kcpp_data = nullptr;
 static int max_context_limit_at_load = 0;
 static int n_past = 0;
 static int debugmode = 0; //-1 = hide all, 0 = normal, 1 = showall
-static bool quiet = false;
+static bool is_quiet = false;
 static std::vector<gpt_vocab::id> last_n_tokens;
 static std::vector<gpt_vocab::id> current_context_tokens;
 static size_t mem_per_token = 0;
@@ -939,12 +939,12 @@ void sample_xtc(llama_token_data_array * candidates, float xtc_threshold, float
 
     if(last_idx>1) //if there are 2 or more viable candidates
     {
-        if (debugmode==1 && !quiet) {
+        if (debugmode==1 && !is_quiet) {
             printf("XTC penalties [");
         }
         // then remove all other tokens above threshold EXCEPT the least likely one
         for (size_t i = 0; i < last_idx - 1; ++i) {
-            if (debugmode==1 && !quiet)
+            if (debugmode==1 && !is_quiet)
             {
                 gpt_vocab::id token = candidates->data[i].id;
                 std::string tokenizedstr = FileFormatTokenizeID(token, file_format);
@@ -953,7 +953,7 @@ void sample_xtc(llama_token_data_array * candidates, float xtc_threshold, float
             }
             candidates->data[i].logit -= 999.0f; //infinity gets wonky results downstream, this hack works well enough
         }
-        if (debugmode==1 && !quiet) {
+        if (debugmode==1 && !is_quiet) {
             printf("]\n");
         }
         candidates->sorted = false;
@@ -1142,7 +1142,7 @@ void sample_dry(int n_ctx, int penalty_range, float penalty_multiplier, float pe
         max_exponent = FLOAT_MAX_LOG / std::log(penalty_base);
     }
 
-    if (debugmode==1 && !quiet && !dry_max_token_repeat.empty()) {
+    if (debugmode==1 && !is_quiet && !dry_max_token_repeat.empty()) {
         printf("DRY penalties [");
     }
     size_t count = 0;
@@ -1153,7 +1153,7 @@ void sample_dry(int n_ctx, int penalty_range, float penalty_multiplier, float pe
             repeat_exp = max_exponent;
         }
         float penalty = penalty_multiplier * pow(penalty_base, repeat_exp);
-        if (debugmode==1 && !quiet)
+        if (debugmode==1 && !is_quiet)
         {
             std::string tokenizedstr = FileFormatTokenizeID(token, file_format);
             ::utreplace(tokenizedstr, "\n", "\\n");
@@ -1166,7 +1166,7 @@ void sample_dry(int n_ctx, int penalty_range, float penalty_multiplier, float pe
     {
         candidates->sorted = false;
     }
-    if (debugmode==1 && !quiet && !dry_max_token_repeat.empty()) {
+    if (debugmode==1 && !is_quiet && !dry_max_token_repeat.empty()) {
         printf("]\n");
     }
 }
@@ -1697,7 +1697,7 @@ static void load_grammar(const std::string & gammarstr)
             printf("\nIgnored invalid grammar sampler.");
             return;
         }
-        if(debugmode==1 && !quiet)
+        if(debugmode==1 && !is_quiet)
         {
             parsed_grammar.print(stderr);
         }
@@ -1840,7 +1840,7 @@ static float CalcGradientAIRopeFreqBase(float original_rope_base, int n_ctx_trai
         float chi_ctx_value = (n_ctx_desired * ctx_multiplier) / 6.28318;
         float gradient_ai_rope_freq_base_value = powf(original_rope_base, log10f(chi_ctx_value) / log10f(chi_ctx_train_value));
 
-        if(debugmode==1 && !quiet)
+        if(debugmode==1 && !is_quiet)
         {
             printf("Trained max context length (value:%.d).\n", n_ctx_train);
             printf("Desired context length (value:%.d).\n", n_ctx_desired);
@@ -1857,7 +1857,7 @@ static float CalcGradientAIRopeFreqBase(float original_rope_base, int n_ctx_trai
         {
             float extended_rope_positive_offset_value = 1 + ((log10f(chi_ctx_value) - log10f(chi_ctx_train_value)) / ((log10f(chi_ctx_value) * log10f(chi_ctx_train_value)) - (log10f(chi_ctx_value) + log10f(chi_ctx_train_value))));
             float rope_freq_base_with_positive_offset = gradient_ai_rope_freq_base_value * extended_rope_positive_offset_value;
-            if(debugmode==1 && !quiet)
+            if(debugmode==1 && !is_quiet)
             {
                 printf("Extended RoPE Positive Offset (multiplicator) for Solar based models. (value:%.3f).\n", extended_rope_positive_offset_value);
                 printf("RoPE base calculated via Gradient AI formula for Solar based models. (value:%.1f).\n", rope_freq_base_with_positive_offset);
@@ -1873,6 +1873,7 @@ static float CalcGradientAIRopeFreqBase(float original_rope_base, int n_ctx_trai
 
 ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in_file_format, FileFormatExtraMeta in_file_format_meta)
 {
+    is_quiet = inputs.quiet;
     ggml_time_init();
     kcpp_data = new kcpp_params(); //allocate on heap to avoid linux segfault. yes this leaks memory.
 
@@ -2688,13 +2689,13 @@ std::vector<int> gpttype_get_token_arr(const std::string & input, bool addbos)
         printf("\nWarning: KCPP text generation not initialized!\n");
         return toks;
     }
-    if(debugmode==1 && !quiet)
+    if(debugmode==1 && !is_quiet)
     {
         printf("\nFileFormat: %d, Tokenizing: %s",file_format ,input.c_str());
     }
     TokenizeString(input, toks, file_format,addbos);
     int tokcount = toks.size();
-    if(debugmode==1 && !quiet)
+    if(debugmode==1 && !is_quiet)
     {
         printf("\nTokens Counted: %d\n",tokcount);
     }
@@ -2779,7 +2780,6 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
         llama_perf_context_reset(llama_ctx_v4);
     }
 
-    quiet = inputs.quiet;
     generation_finished = false; // Set current generation status
     generated_tokens.clear(); // New Generation, new tokens
     delayed_generated_tokens.clear();
@@ -2858,7 +2858,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
     banned_token_ids.clear();
     if(banned_tokens.size()>0)
     {
-        if(debugmode==1 && !quiet)
+        if(debugmode==1 && !is_quiet)
         {
             printf("\nBanning %zu single character sequences...",banned_tokens.size());
         }
@@ -2875,13 +2875,13 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
                 }
             }
         }
-        if(debugmode==1 && !quiet)
+        if(debugmode==1 && !is_quiet)
         {
             printf("\nBanned a total of %zu individual tokens.\n",banned_token_ids.size());
         }
     }
 
-    if(debugmode==1 && !quiet && banned_phrases.size()>0)
+    if(debugmode==1 && !is_quiet && banned_phrases.size()>0)
     {
         printf("\nBanned a total of %zu phrases, with max token count of %d.\n",banned_phrases.size(),delayed_generated_tokens_limit);
     }
@@ -2926,7 +2926,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
         //images have changed. swap identifiers to force reprocessing
         current_llava_identifier = (current_llava_identifier==LLAVA_TOKEN_IDENTIFIER_A?LLAVA_TOKEN_IDENTIFIER_B:LLAVA_TOKEN_IDENTIFIER_A);
         llava_composite_image_signature = new_llava_composite;
-        if(debugmode==1 && !quiet)
+        if(debugmode==1 && !is_quiet)
         {
             printf("\nLLAVA images changed, existing cache invalidated");
         }
@@ -2982,7 +2982,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
             const int MAX_CHAR_LEN = 40;
             const int MAX_SEQ_LEN = 20;
 
-            if (debugmode == 1 && !quiet)
+            if (debugmode == 1 && !is_quiet)
             {
                 printf("\nProcessing %zu dry break strings...", kcpp_data->dry_sequence_breakers.size());
             }
@@ -2994,7 +2994,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
                 }
                 GetOverlappingTokenSequences(sequence_break, dry_sequence_breakers, MAX_SEQ_LEN);
             }
-            if (debugmode == 1 && !quiet)
+            if (debugmode == 1 && !is_quiet)
             {
                 int trivial = 0, non_trivial = 0;
                 for (const auto &seq : dry_sequence_breakers)
@@ -3014,7 +3014,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
     }
 
     bool stream_sse = inputs.stream_sse;
-    bool allow_regular_prints = (!quiet && debugmode!=-1);
+    bool allow_regular_prints = (!is_quiet && debugmode!=-1);
 
     std::string grammarstr = inputs.grammar;
     bool grammar_retain_state = inputs.grammar_retain_state;
@@ -3047,7 +3047,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
     if (kcpp_data->seed <= 0 || kcpp_data->seed==0xFFFFFFFF)
     {
         kcpp_data->seed = (((uint32_t)time(NULL)) % 1000000u);
-        if(debugmode==1 && !quiet)
+        if(debugmode==1 && !is_quiet)
         {
             printf("\nUsing Seed: %d",kcpp_data->seed);
         }
@@ -3079,7 +3079,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
             }
             else
             {
-                if(debugmode==1 && !quiet)
+                if(debugmode==1 && !is_quiet)
                 {
                     printf("\nCreating clip image embed...");
                 }
@@ -3087,7 +3087,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
                 if (!llava_image_embed_make_with_clip_img(clp_ctx, kcpp_data->n_threads, clp_img_data, &llava_images[i].clp_img_embd, &llava_images[i].clp_image_tokens)) {
                     printf("\nError: Clip image %d failed to create embd!",i);
                 }
-                if(debugmode==1 && !quiet)
+                if(debugmode==1 && !is_quiet)
                 {
                     printf("\nLLAVA Clip Embed %i used Tokens: %d",i,llava_images[i].clp_image_tokens);
                 }
@@ -3210,7 +3210,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
     std::fill(last_n_tokens.begin(), last_n_tokens.end(), 0);
     n_past = 0;
 
-    if (debugmode==1 && !quiet)
+    if (debugmode==1 && !is_quiet)
     {
         std::string outstr = "";
         printf("\n\n[Debug: Dump Raw Input Tokens, format: %d]\n", file_format);
@@ -3355,7 +3355,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
         printf("\n");
     }
 
-    if (debugmode==1 && !quiet)
+    if (debugmode==1 && !is_quiet)
     {
         std::string outstr = "";
         printf("\n[Debug: Dump Forwarded Input Tokens, format: %d]\n", file_format);
@@ -3404,7 +3404,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
                     draft_used = true;
                     draft_results = speculative_decoding_eval_chunk(draft_ctx, llama_ctx_v4, embd, n_vocab, n_past);
                     evalres = draft_results.draft_success;
-                    if(debugmode==1 && !quiet)
+                    if(debugmode==1 && !is_quiet)
                     {
                         std::string draftedtoks = get_tok_vec_str(draft_results.draftids);
                         printf("\nDrafted %d Tokens: [%s]\n",speculative_chunk_amt,draftedtoks.c_str());
@@ -3607,7 +3607,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
                 if(draft_used)
                 {
                     int32_t draftedid = draft_results.draftids[logits_sampled];
-                    if(debugmode==1 && !quiet)
+                    if(debugmode==1 && !is_quiet)
                     {
                         std::string drafttok = FileFormatTokenizeID(draftedid, file_format, true);
                         std::string realtok = FileFormatTokenizeID(id, file_format, true);
@@ -3660,7 +3660,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
                 {
                     printf("\rGenerating (%d / %d tokens)", (kcpp_data->n_predict - remaining_tokens), kcpp_data->n_predict);
                 }
-                if(debugmode==1 && !quiet && top_picks_history.size()>0)
+                if(debugmode==1 && !is_quiet && top_picks_history.size()>0)
                 {
                     printf(" [");
                     bool firstloop = true;
@@ -3912,7 +3912,7 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
         delayed_generated_tokens.pop_front();
     }
 
-    if(debugmode==1 && !quiet && file_format == FileFormat::GGUF_GENERIC)
+    if(debugmode==1 && !is_quiet && file_format == FileFormat::GGUF_GENERIC)
     {
         printf("\n");
         llama_perf_context_print(llama_ctx_v4);
diff --git a/klite.embd b/klite.embd
index a583f5a0d..f6c6b0b5d 100644
--- a/klite.embd
+++ b/klite.embd
@@ -12,7 +12,7 @@ Current version indicated by LITEVER below.
 -->
 
 <script>
-	const LITEVER = 205;
+	const LITEVER = 206;
 	const urlParams = new URLSearchParams(window.location.search);
 	var localflag = true;
 	const STORAGE_PREFIX = (localflag?"e_":"")+"kaihordewebui_";
@@ -5009,8 +5009,17 @@ initializeInstructUIFunctionality();
 	function copyMarkdownCode(btn)
 	{
 		const codeContainer = btn.parentElement.querySelector('pre code');
-		//selectElementContents(codeContainer);
-		navigator.clipboard.writeText(codeContainer.innerText);
+		let innercode = codeContainer.innerText;
+		//remove common language descriptiors from the start
+		let langsmatched = ["matlab","jsonc","powershell","ps1","haskell","hs","vbnet","vb","apache","apacheconf","makefile","mk","ini","protobuf","proto","typescript","tsx","markdown","md","mkdown","mkd","python","py","javascript","js","jsx","html","xhtml","xml","css","json","typescript","ts","tsx","bash","sh","zsh","java","csharp","cs","c","h","cpp","hpp","php","sql","ruby","rb","go","golang","kotlin","kt","swift","rust","rs","r","dart","scala","dockerfile","docker","yaml","yml","ini","toml","perl","pl","shell","console","powershell","ps1","lua","typescript","ts"];
+		for(let i = 0; i < langsmatched.length; ++i) {
+			let matcher = langsmatched[i]+"\n";
+			if (innercode.startsWith(matcher)) {
+				innercode = innercode.substring(matcher.length);
+				break;
+			 }
+		}
+		navigator.clipboard.writeText(innercode);
 	}
 
 	function simpleMarkdown(text) {
@@ -13469,7 +13478,12 @@ initializeInstructUIFunctionality();
 					if (document.getElementById("jailbreakprompt2") && document.getElementById("jailbreakprompt2").checked && document.getElementById("jailbreakprompttext2").value!="") {
 						let addrole = document.getElementById("jailbreakprompttext2role").value;
 						addrole = ((addrole==2)?"system":(addrole==1?"assistant":"user"));
-						oai_payload.messages.push({ "role": addrole, "content": document.getElementById("jailbreakprompttext2").value });
+						let postmsg = { "role": addrole, "content": document.getElementById("jailbreakprompttext2").value };
+						if(addrole=="assistant" && targetep.toLowerCase().includes("api.deepseek.com"))
+						{
+							postmsg["prefix"] = true;
+						}
+						oai_payload.messages.push(postmsg);
 					}
 
 					oaiemulatecompletionscontent = "";
diff --git a/koboldcpp.py b/koboldcpp.py
index 921916058..1745abc78 100644
--- a/koboldcpp.py
+++ b/koboldcpp.py
@@ -160,7 +160,6 @@ class load_model_inputs(ctypes.Structure):
                 ("cublas_info", ctypes.c_int),
                 ("vulkan_info", ctypes.c_char_p),
                 ("blasbatchsize", ctypes.c_int),
-                ("debugmode", ctypes.c_int),
                 ("forceversion", ctypes.c_int),
                 ("gpulayers", ctypes.c_int),
                 ("rope_freq_scale", ctypes.c_float),
@@ -169,7 +168,9 @@ class load_model_inputs(ctypes.Structure):
                 ("flash_attention", ctypes.c_bool),
                 ("tensor_split", ctypes.c_float * tensor_split_max),
                 ("quant_k", ctypes.c_int),
-                ("quant_v", ctypes.c_int)]
+                ("quant_v", ctypes.c_int),
+                ("quiet", ctypes.c_bool),
+                ("debugmode", ctypes.c_int)]
 
 class generation_inputs(ctypes.Structure):
     _fields_ = [("seed", ctypes.c_int),
@@ -202,7 +203,6 @@ class generation_inputs(ctypes.Structure):
                 ("stream_sse", ctypes.c_bool),
                 ("grammar", ctypes.c_char_p),
                 ("grammar_retain_state", ctypes.c_bool),
-                ("quiet", ctypes.c_bool),
                 ("dynatemp_range", ctypes.c_float),
                 ("dynatemp_exponent", ctypes.c_float),
                 ("smoothing_factor", ctypes.c_float),
@@ -242,6 +242,7 @@ class sd_load_model_inputs(ctypes.Structure):
                 ("vae_filename", ctypes.c_char_p),
                 ("lora_filename", ctypes.c_char_p),
                 ("lora_multiplier", ctypes.c_float),
+                ("quiet", ctypes.c_bool),
                 ("debugmode", ctypes.c_int)]
 
 class sd_generation_inputs(ctypes.Structure):
@@ -255,8 +256,7 @@ class sd_generation_inputs(ctypes.Structure):
                 ("height", ctypes.c_int),
                 ("seed", ctypes.c_int),
                 ("sample_method", ctypes.c_char_p),
-                ("clip_skip", ctypes.c_int),
-                ("quiet", ctypes.c_bool)]
+                ("clip_skip", ctypes.c_int)]
 
 class sd_generation_outputs(ctypes.Structure):
     _fields_ = [("status", ctypes.c_int),
@@ -268,14 +268,14 @@ class whisper_load_model_inputs(ctypes.Structure):
                 ("clblast_info", ctypes.c_int),
                 ("cublas_info", ctypes.c_int),
                 ("vulkan_info", ctypes.c_char_p),
+                ("quiet", ctypes.c_bool),
                 ("debugmode", ctypes.c_int)]
 
 class whisper_generation_inputs(ctypes.Structure):
     _fields_ = [("prompt", ctypes.c_char_p),
                 ("audio_data", ctypes.c_char_p),
                 ("suppress_non_speech", ctypes.c_bool),
-                ("langcode", ctypes.c_char_p),
-                ("quiet", ctypes.c_bool)]
+                ("langcode", ctypes.c_char_p)]
 
 class whisper_generation_outputs(ctypes.Structure):
     _fields_ = [("status", ctypes.c_int),
@@ -291,13 +291,13 @@ class tts_load_model_inputs(ctypes.Structure):
                 ("vulkan_info", ctypes.c_char_p),
                 ("gpulayers", ctypes.c_int),
                 ("flash_attention", ctypes.c_bool),
+                ("quiet", ctypes.c_bool),
                 ("debugmode", ctypes.c_int)]
 
 class tts_generation_inputs(ctypes.Structure):
     _fields_ = [("prompt", ctypes.c_char_p),
                 ("speaker_seed", ctypes.c_int),
                 ("audio_seed", ctypes.c_int),
-                ("quiet", ctypes.c_bool),
                 ("nocache", ctypes.c_bool)]
 
 class tts_generation_outputs(ctypes.Structure):
@@ -513,6 +513,12 @@ def set_backend_props(inputs):
         inputs.vulkan_info = s.encode("UTF-8")
     else:
         inputs.vulkan_info = "".encode("UTF-8")
+
+    # set universal flags
+    inputs.quiet = args.quiet
+    inputs.debugmode = args.debugmode
+    inputs.executable_path = (getdirpath()+"/").encode("UTF-8")
+
     return inputs
 
 def end_trim_to_sentence(input_text):
@@ -1077,13 +1083,10 @@ def load_model(model_filename):
 
     inputs.moe_experts = args.moeexperts
     inputs = set_backend_props(inputs)
-
-    inputs.executable_path = (getdirpath()+"/").encode("UTF-8")
-    inputs.debugmode = args.debugmode
     ret = handle.load_model(inputs)
     return ret
 
-def generate(genparams, is_quiet=False, stream_flag=False):
+def generate(genparams, stream_flag=False):
     global maxctx, args, currentusergenkey, totalgens, pendingabortkey
 
     prompt = genparams.get('prompt', "")
@@ -1121,7 +1124,6 @@ def generate(genparams, is_quiet=False, stream_flag=False):
     grammar_retain_state = genparams.get('grammar_retain_state', False)
     genkey = genparams.get('genkey', '')
     trimstop = genparams.get('trim_stop', True)
-    quiet = is_quiet
     dynatemp_range = genparams.get('dynatemp_range', 0.0)
     dynatemp_exponent = genparams.get('dynatemp_exponent', 1.0)
     smoothing_factor = genparams.get('smoothing_factor', 0.0)
@@ -1170,7 +1172,6 @@ def generate(genparams, is_quiet=False, stream_flag=False):
     inputs.rep_pen_slope = rep_pen_slope
     inputs.presence_penalty = presence_penalty
     inputs.stream_sse = stream_sse
-    inputs.quiet = quiet
     inputs.dynatemp_range = dynatemp_range
     inputs.dynatemp_exponent = dynatemp_exponent
     inputs.smoothing_factor = smoothing_factor
@@ -1289,8 +1290,6 @@ def generate(genparams, is_quiet=False, stream_flag=False):
 def sd_load_model(model_filename,vae_filename,lora_filename,t5xxl_filename,clipl_filename,clipg_filename):
     global args
     inputs = sd_load_model_inputs()
-    inputs.debugmode = args.debugmode
-    inputs.executable_path = (getdirpath()+"/").encode("UTF-8")
     inputs.model_filename = model_filename.encode("UTF-8")
     thds = args.threads
     quant = 0
@@ -1368,7 +1367,6 @@ def sd_generate(genparams):
     height = tryparseint(genparams.get("height", 512))
     seed = tryparseint(genparams.get("seed", -1))
     sample_method = genparams.get("sampler_name", "k_euler_a")
-    is_quiet = True if (args.quiet or args.debugmode == -1) else False
     clip_skip = tryparseint(genparams.get("clip_skip", -1))
 
     #clean vars
@@ -1405,7 +1403,6 @@ def sd_generate(genparams):
     inputs.height = height
     inputs.seed = seed
     inputs.sample_method = sample_method.lower().encode("UTF-8")
-    inputs.quiet = is_quiet
     inputs.clip_skip = clip_skip
     ret = handle.sd_generate(inputs)
     outstr = ""
@@ -1417,8 +1414,6 @@ def sd_generate(genparams):
 def whisper_load_model(model_filename):
     global args
     inputs = whisper_load_model_inputs()
-    inputs.debugmode = args.debugmode
-    inputs.executable_path = (getdirpath()+"/").encode("UTF-8")
     inputs.model_filename = model_filename.encode("UTF-8")
     inputs = set_backend_props(inputs)
     ret = handle.whisper_load_model(inputs)
@@ -1426,7 +1421,6 @@ def whisper_load_model(model_filename):
 
 def whisper_generate(genparams):
     global args
-    is_quiet = True if (args.quiet or args.debugmode == -1) else False
     prompt = genparams.get("prompt", "")
     audio_data = genparams.get("audio_data", "")
     if audio_data.startswith("data:audio"):
@@ -1434,7 +1428,6 @@ def whisper_generate(genparams):
     inputs = whisper_generation_inputs()
     inputs.prompt = prompt.encode("UTF-8")
     inputs.audio_data = audio_data.encode("UTF-8")
-    inputs.quiet = is_quiet
     lc = genparams.get("langcode", genparams.get("language", "auto"))
     lc = lc.strip().lower() if (lc and lc.strip().lower()!="") else "auto"
     inputs.langcode = lc.encode("UTF-8")
@@ -1448,8 +1441,6 @@ def whisper_generate(genparams):
 def tts_load_model(ttc_model_filename,cts_model_filename):
     global args
     inputs = tts_load_model_inputs()
-    inputs.debugmode = args.debugmode
-    inputs.executable_path = (getdirpath()+"/").encode("UTF-8")
     inputs.ttc_model_filename = ttc_model_filename.encode("UTF-8")
     inputs.cts_model_filename = cts_model_filename.encode("UTF-8")
     inputs.gpulayers = (999 if args.ttsgpu else 0)
@@ -1466,7 +1457,6 @@ def tts_load_model(ttc_model_filename,cts_model_filename):
 
 def tts_generate(genparams):
     global args
-    is_quiet = True if (args.quiet or args.debugmode == -1) else False
     prompt = genparams.get("input", genparams.get("text", ""))
     prompt = prompt.strip()
     voice = 1
@@ -1486,7 +1476,6 @@ def tts_generate(genparams):
     except Exception:
         aseed = -1
     inputs.audio_seed = aseed
-    inputs.quiet = is_quiet
     inputs.nocache = genparams.get("nocache", False)
     ret = handle.tts_generate(inputs)
     outstr = ""
@@ -2044,7 +2033,6 @@ class ServerRequestHandler(http.server.SimpleHTTPRequestHandler):
 
     async def generate_text(self, genparams, api_format, stream_flag):
         global friendlymodelname, chatcompl_adapter, currfinishreason
-        is_quiet = args.quiet
         currfinishreason = "null"
 
         def run_blocking():  # api format 1=basic,2=kai,3=oai,4=oai-chat
@@ -2054,7 +2042,7 @@ class ServerRequestHandler(http.server.SimpleHTTPRequestHandler):
                 global last_non_horde_req_time
                 last_non_horde_req_time = time.time()
 
-            return generate(genparams=genparams,is_quiet=is_quiet,stream_flag=stream_flag)
+            return generate(genparams=genparams,stream_flag=stream_flag)
 
         genout = {"text": "", "status": -1, "stopreason": -1, "prompt_tokens":0, "completion_tokens": 0, "total_tokens": 0}
         if stream_flag:
diff --git a/otherarch/sdcpp/sdtype_adapter.cpp b/otherarch/sdcpp/sdtype_adapter.cpp
index 687f1589d..58c5bc6e9 100644
--- a/otherarch/sdcpp/sdtype_adapter.cpp
+++ b/otherarch/sdcpp/sdtype_adapter.cpp
@@ -114,8 +114,11 @@ static std::string recent_data = "";
 
 static std::string sdplatformenv, sddeviceenv, sdvulkandeviceenv;
 static bool notiling = false;
-bool sdtype_load_model(const sd_load_model_inputs inputs) {
+static bool sd_is_quiet = false;
 
+bool sdtype_load_model(const sd_load_model_inputs inputs) {
+    sd_is_quiet = inputs.quiet;
+    set_sd_quiet(sd_is_quiet);
     executable_path = inputs.executable_path;
     std::string taesdpath = "";
     std::string lorafilename = inputs.lora_filename;
@@ -290,9 +293,6 @@ sd_generation_outputs sdtype_generate(const sd_generation_inputs inputs)
     sd_image_t * results;
     sd_image_t* control_image = NULL;
 
-    bool is_quiet = inputs.quiet;
-    set_sd_quiet(is_quiet);
-
     //sanitize prompts, remove quotes and limit lengths
     std::string cleanprompt = clean_input_prompt(inputs.prompt);
     std::string cleannegprompt = clean_input_prompt(inputs.negative_prompt);
@@ -345,7 +345,7 @@ sd_generation_outputs sdtype_generate(const sd_generation_inputs inputs)
     std::vector<uint8_t> resized_image_buf(img2imgW * img2imgH * img2imgC);
 
     std::string ts = get_timestamp_str();
-    if(!is_quiet)
+    if(!sd_is_quiet)
     {
         printf("\n[%s] Generating Image (%d steps)\n",ts.c_str(),inputs.sample_steps);
     }else{
@@ -385,7 +385,7 @@ sd_generation_outputs sdtype_generate(const sd_generation_inputs inputs)
 
     if (sd_params->mode == TXT2IMG) {
 
-        if(!is_quiet && sddebugmode==1)
+        if(!sd_is_quiet && sddebugmode==1)
         {
             printf("\nTXT2IMG PROMPT:%s\nNPROMPT:%s\nCLPSKP:%d\nCFGSCLE:%f\nW:%d\nH:%d\nSM:%d\nSTEP:%d\nSEED:%d\nBATCH:%d\nCIMG:%p\nCSTR:%f\n\n",
             sd_params->prompt.c_str(),
@@ -471,7 +471,7 @@ sd_generation_outputs sdtype_generate(const sd_generation_inputs inputs)
         input_image.channel = img2imgC;
         input_image.data = resized_image_buf.data();
 
-        if(!is_quiet && sddebugmode==1)
+        if(!sd_is_quiet && sddebugmode==1)
         {
             printf("\nIMG2IMG PROMPT:%s\nNPROMPT:%s\nCLPSKP:%d\nCFGSCLE:%f\nW:%d\nH:%d\nSM:%d\nSTEP:%d\nSEED:%d\nBATCH:%d\nCIMG:%p\nSTR:%f\n\n",
             sd_params->prompt.c_str(),
diff --git a/otherarch/tts_adapter.cpp b/otherarch/tts_adapter.cpp
index 897f8e401..bd405acd4 100644
--- a/otherarch/tts_adapter.cpp
+++ b/otherarch/tts_adapter.cpp
@@ -466,6 +466,7 @@ static llama_context * cts_ctx = nullptr; //codes to speech
 
 static TTS_VER ttsver = TTS_VER_2;
 static int ttsdebugmode = 0;
+static bool tts_is_quiet = false;
 static std::string ttsplatformenv, ttsdeviceenv, ttsvulkandeviceenv;
 static std::string last_generated_audio = "";
 static std::string last_generation_settings_prompt = ""; //for caching purposes to fix ST bug
@@ -480,6 +481,8 @@ static int nthreads = 4;
 
 bool ttstype_load_model(const tts_load_model_inputs inputs)
 {
+    tts_is_quiet = inputs.quiet;
+
     //duplicated from expose.cpp
     int cl_parseinfo = inputs.clblast_info; //first digit is whether configured, second is platform, third is devices
     std::string usingclblast = "GGML_OPENCL_CONFIGURED="+std::to_string(cl_parseinfo>0?1:0);
@@ -623,7 +626,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
     {
         audio_seed = (((uint32_t)time(NULL)) % 1000000u);
     }
-    if(ttsdebugmode==1 && !inputs.quiet)
+    if(ttsdebugmode==1 && !tts_is_quiet)
     {
         printf("\nUsing Speaker Seed: %d", speaker_seed);
         printf("\nUsing Audio Seed: %d", audio_seed);
@@ -639,7 +642,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
     && last_generated_audio!=""
     && last_generation_settings_prompt == std::string(inputs.prompt))
     {
-        if (ttsdebugmode == 1 && !inputs.quiet) {
+        if (ttsdebugmode == 1 && !tts_is_quiet) {
             printf("\nReusing Cached Audio.\n");
         }
         output.data = last_generated_audio.c_str();
@@ -662,7 +665,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
     if(empty_check)
     {
         //no input
-         if(!inputs.quiet)
+         if(!tts_is_quiet)
         {
             printf("\nTTS sent empty input.\n");
             last_generated_audio = "";
@@ -676,7 +679,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
     timer_start();
 
 
-    if(!inputs.quiet && ttsdebugmode==1)
+    if(!tts_is_quiet && ttsdebugmode==1)
     {
         printf("\nInput: %s\n", prompt_clean.c_str());
     }
@@ -691,7 +694,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
         if(last_speaker_seed==speaker_seed && !last_speaker_codes.empty())
         {
             //able to proceed, do nothing
-            if(!inputs.quiet && ttsdebugmode==1)
+            if(!tts_is_quiet && ttsdebugmode==1)
             {
                 printf("\nReuse speaker ID=%d (%d tokens)...", last_speaker_seed, last_speaker_codes.size());
             }
@@ -717,7 +720,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
             }
             last_speaker_codes = common_tokenize(ttcvocab, speaker, false, true);
             last_speaker_seed = speaker_seed;
-            if(!inputs.quiet && ttsdebugmode==1)
+            if(!tts_is_quiet && ttsdebugmode==1)
             {
                 printf("\nSpecial ID=%d (%d tokens)...", last_speaker_seed, last_speaker_codes.size());
             }
@@ -725,7 +728,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
             //generate the voice texture of our new speaker
             last_speaker_codes.clear();
             guide_tokens = prepare_guide_tokens(ttcvocab,sampletext,ttsver);
-            if(!inputs.quiet && ttsdebugmode==1)
+            if(!tts_is_quiet && ttsdebugmode==1)
             {
                 printf("\nGuide Tokens (%d tokens):\n", guide_tokens.size());
                 const std::string inp_txt = common_detokenize(ttc_ctx, guide_tokens, true);
@@ -734,7 +737,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
             }
             prompt_add(prompt_inp, ttcvocab, sampletext, false, true);
             prompt_add(prompt_inp, ttcvocab, "<|text_end|>\n<|audio_start|>\n", false, true);
-            if(!inputs.quiet && ttsdebugmode==1)
+            if(!tts_is_quiet && ttsdebugmode==1)
             {
                 printf("\nPrepare new speaker (%d input tokens)...\n", prompt_inp.size());
                 print_tok_vec(prompt_inp);
@@ -806,7 +809,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
                 }
             }
             last_speaker_seed = speaker_seed;
-            if(!inputs.quiet && ttsdebugmode==1)
+            if(!tts_is_quiet && ttsdebugmode==1)
             {
                 printf("\nNew speaker ID=%d created (%d tokens)...", last_speaker_seed, last_speaker_codes.size());
                 const std::string inp_txt = common_detokenize(ttc_ctx, last_speaker_codes, true);
@@ -821,7 +824,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
 
     //second pass: add the speaker before the actual prompt
     guide_tokens = prepare_guide_tokens(ttcvocab,prompt_clean,ttsver);
-    if(!inputs.quiet && ttsdebugmode==1)
+    if(!tts_is_quiet && ttsdebugmode==1)
     {
         printf("\nGuide Tokens (%d tokens):\n", guide_tokens.size());
         const std::string inp_txt = common_detokenize(ttc_ctx, guide_tokens, true);
@@ -834,7 +837,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
     }
     prompt_add(prompt_inp, ttcvocab, prompt_clean, false, true);
 
-    if(!inputs.quiet)
+    if(!tts_is_quiet)
     {
         printf("\nTTS Processing (%d input tokens)...\n", prompt_inp.size());
     }
@@ -847,7 +850,7 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
         prompt_add(prompt_inp, ttcvocab, "\n", false, true);
     }
 
-    if(!inputs.quiet && ttsdebugmode==1)
+    if(!tts_is_quiet && ttsdebugmode==1)
     {
         printf("\nDUMP TTS PROMPT (%d tokens):\n", prompt_inp.size());
         print_tok_vec(prompt_inp);
@@ -913,13 +916,13 @@ tts_generation_outputs ttstype_generate(const tts_generation_inputs inputs)
             output.status = 0;
             return output;
         }
-        if(!inputs.quiet)
+        if(!tts_is_quiet)
         {
             printf("\rTTS Generating (%d outputs)", n_decode);
         }
     }
 
-    if(!inputs.quiet && ttsdebugmode==1)
+    if(!tts_is_quiet && ttsdebugmode==1)
     {
         const std::string inp_txt = common_detokenize(ttc_ctx, codes, true);
         printf("\nGenerated %d Codes: '%s'\n",codes.size(), inp_txt.c_str());
diff --git a/otherarch/whispercpp/whisper_adapter.cpp b/otherarch/whispercpp/whisper_adapter.cpp
index 9aa20603f..f839576e2 100644
--- a/otherarch/whispercpp/whisper_adapter.cpp
+++ b/otherarch/whispercpp/whisper_adapter.cpp
@@ -24,7 +24,7 @@
 #endif
 
 static int whisperdebugmode = 0;
-static bool whisperquiet = false;
+static bool whisper_is_quiet = false;
 static whisper_context * whisper_ctx = nullptr;
 static std::string whisper_output_text = "";
 
@@ -90,7 +90,7 @@ static bool read_wav(const std::string & b64data, std::vector<float>& pcmf32, st
     std::vector<float> raw_pcm;
     raw_pcm.resize(n);
 
-    if(whisperdebugmode==1 && !whisperquiet)
+    if(whisperdebugmode==1 && !whisper_is_quiet)
     {
         printf("\nwav_data_size: %d, n:%d",wav_data.size(),n);
     }
@@ -107,7 +107,7 @@ static bool read_wav(const std::string & b64data, std::vector<float>& pcmf32, st
     }
 
     if (wav.sampleRate != COMMON_SAMPLE_RATE) {
-        if(whisperdebugmode==1 && !whisperquiet)
+        if(whisperdebugmode==1 && !whisper_is_quiet)
         {
             printf("\nResample wav from %" PRIu32 " to %" PRIu32 " (in size: %zu)",
             wav.sampleRate, COMMON_SAMPLE_RATE, raw_pcm.size());
@@ -140,6 +140,8 @@ void cb_log_disable(enum ggml_log_level , const char * , void * ) { }
 static std::string whisperplatformenv, whisperdeviceenv, whispervulkandeviceenv;
 bool whispertype_load_model(const whisper_load_model_inputs inputs)
 {
+    whisper_is_quiet = inputs.quiet;
+
     //duplicated from expose.cpp
     int cl_parseinfo = inputs.clblast_info; //first digit is whether configured, second is platform, third is devices
     std::string usingclblast = "GGML_OPENCL_CONFIGURED="+std::to_string(cl_parseinfo>0?1:0);
@@ -203,8 +205,7 @@ whisper_generation_outputs whispertype_generate(const whisper_generation_inputs
         return output;
     }
 
-    whisperquiet = inputs.quiet;
-    if(!whisperquiet)
+    if(!whisper_is_quiet)
     {
         printf("\nWhisper Transcribe Generating...");
     }
@@ -263,14 +264,14 @@ whisper_generation_outputs whispertype_generate(const whisper_generation_inputs
         return output;
     }
 
-    if (!whisperquiet && whisperdebugmode==1) {
+    if (!whisper_is_quiet && whisperdebugmode==1) {
         whisper_print_timings(whisper_ctx);
     }
 
     // output text transcription
     whisper_output_text = output_txt(whisper_ctx, pcmf32s);
     std::string ts = get_timestamp_str();
-    if(!whisperquiet)
+    if(!whisper_is_quiet)
     {
         printf("\n[%s] Whisper Transcribe Output: %s",ts.c_str(),whisper_output_text.c_str());
     } else {